Linklaters builds Nakhoda platform to reinvent contracting for AI-ready legal documents

Artificial Lawyer's February 2019 profile reports the Nakhoda team's view that AI systems, as we have them today, "don't really work". Its analogy is a search that gives "70% of the result you need" and then tells you to finish the job by hand. They put the cause in the documents: "legal documents are often unpredictable mish-mashes of unstructured data", and they argue that the only way to solve the legal data problem is to solve it at its point of creation.

Linklaters has its own legal AI platform, Nakhoda, built by an in-house team that has been developing steadily since 2015 and today numbers around 25 people. Rather than extracting data from legacy documents, the approach is to create contracts in a digital environment using document automation, templates, model clauses and an agreed methodology, so documents and their legal data are highly predictable for algorithms from the moment they are created. One example is a Nakhoda platform service for ISDA contracts, which Linklaters is developing with the global derivatives rules body for a wave of new rules on margins; it can also find counterparties to agreements.

The article reports no measured results. It describes the team's hypothesis that automated review would be significantly improved for contracts created this way, because the extraction technology already has a very good idea of what it is about to find. It calls the ISDA system very comprehensive, and says the challenge now is scale and providing for multiple use cases.

• The Nakhoda team's position is that the legal data problem has to be solved where contracts are created, not at extraction.
• Its example is ISDA derivatives contracts; the article calls derivatives an excellent place to start.
• Adoption is the stated open question: "Can we convince people to do something that changes how they do things?"